User guide
Choose a model
Practical guidance for using NeonAPI.
Match the model to the endpoint
Copy the exact ID from Models & pricing. Names in a client preset may differ from the IDs published by the gateway.
| Workflow | Look for |
|---|---|
| Chat and assistants | Chat Completions or Responses support |
| Claude Code | Anthropic Messages support |
| Gemini-native client | Gemini generateContent support |
| Images, speech and video | The corresponding modality and task endpoint |
| Search and retrieval | Embeddings or rerank support |
Evaluate quality and cost
Compare input, output, cache, context and per-task prices where applicable. A lower input rate alone does not guarantee a cheaper workload. Context and output limits also affect whether your request can run.
Check availability
A published model may still be unavailable to a particular key or group. List models with your configured application credentials when diagnosing access. Do not substitute an unlisted model ID or assume every model accepts every protocol.